Choosing the Backhaul for LTE

The advent of LTE trials by major carriers globally leads us to question once again the choice of backhaul technology given the ever increasing bandwidth demands of the new base stations. LTE proponents are predicting bandwidth capabilities per base station in the range of 100 to 300 Mbps. This of course depends on the operator having adequate radio access spectrum to deliver this amount of bandwidth to the end user. In many instances due to limited access spectrum the available throughput will in fact be closer to the lower end of the scale. Nonetheless this represents a significant increase over the existing 3G base stations and will require new backhaul strategies in order to deliver cost effective networks.

 

One viewpoint is that the technology of choice must be fiber, since it can deliver almost unlimited amounts of bandwidth. While it is true that if the mobile operator has already installed a fiber connection between the cell tower and the mobile switching center, then this will be the logical choice for the backhaul. If however, the operator does not own the fiber connection but must lease it from a third party, or must build out a fiber lateral from his existing fiber plant, then the answer is much less certain and alternative technologies such as microwave become much more attractive.

 

In order to understand why the operator might choose one technology over another, let us look at some typical network characteristics. The initial LTE deployments will likely be in the greater metro areas — urban and suburban coverage of the major cities in a large area “hot zone” approach. The average connection length in such a network is driven by the base station density — which in turn is driven by a combination of factors including subscriber density, bandwidth per user, amount of spectrum available to the operator, the spectral efficiency and propagation characteristics of the radio access network. Many of these factors balance out and we can say with some degree of confidence that in a typical North American metro deployment the average connection length will be between 10,000 and 20,000 feet.  Fiber installation costs are between $10 to $40 per foot for rural/suburban areas and between $50 to $200 per foot in dense urban areas. Even taking the low end of the scale for cost and length the fiber build out cost per site would be on the order of $300,000. This is even before any electronics are placed at the end of the fiber to provide capacity. A microwave link, including hardware, installation and an allocation for operations cost over the life of the link would be ~10 times lower cost. The decision seems obvious.

 

But you say there are very cheap leased alternatives from the cable operators and others. While it is true that the cost per bit of these leased Ethernet services is significantly lower when compared to the present mode of operator which is leased T1s, the picture is different when compared to a microwave installation. The second factor is that the base station capacity (100 Mbps) is significantly lower than the capacity of the fiber. The cost per bit of a leased service is driven by the capacity required. At 1 Gbps the cost of a leased fiber Ethernet service is at its lowest (~$5/Mbps/month), however at 100 Mbps even though the monthly charge is lower the cost per Mbps per month is almost triple (~$15/Mbps/month). In addition, there is usually a connection fee to offset the cost of building the fiber lateral to connect the site which increases the effective cost per month by another 35% (~$20/Mbps/Month). This cost is 3 to 5 times higher than the equivalent cost of a packet microwave installation. In addition, the SLA that is available from the cable operator is often not adequate for mobile backhaul in terms of availability or quality of service so even if the costs were significantly lower the performance may not be adequate.

 

With this argument, microwave is more economically attractive, one asks can it deliver the capacities required to service the base stations? Modern packet radios can deliver almost 400 Mbps in a 50 MHz channel (typical of the channels used in North America for backhaul links). Solutions using multiple radio heads or multi carrier techniques can increase this to 1.6 Gbps per link. The capacity advances do not stop there, however. Advances in modulation technology and signal processing promise to almost double this capacity again in the near future. This capacity is clearly adequate for single base station sites, and even for the aggregation of the traffic from multiple base stations.

 

There is no doubt that there will be a mix of both fiber and microwave in any 4G network deployment, whether LTE or WiMAX. The split between the two technologies will depend not only on the economics discussed above, but the strategic imperatives of the operator and the availability of existing fiber based resources. In any case it seems clear that even the most fiber centric carriers will have to take a long hard look at a significant percentage of packet microwave in their 4G networks.